Introduction
Although overall rates of traumatic brain injury (TBI) in adolescents have decreased (Asemota, George, Bowman, Haider, & Schneider, Reference Asemota, George, Bowman, Haider and Schneider2013), incidence changes depend on age (Faul, Xu, Wald, & Coronado, Reference Faul, Xu, Wald and Coronado2010). For example, hospitalizations for falls and bicycle accidents in older adolescents have increased (Asemota et al., Reference Asemota, George, Bowman, Haider and Schneider2013). TBI often results in emotional processing deficits that are devastating to long-term prospects for recovery. Impairments in recognizing (Schmidt, Hanten, Li, Orsten, & Levin, Reference Schmidt, Hanten, Li, Orsten and Levin2010; Tonks et al., Reference Tonks, Williams, Frampton, Yates, Wall and Slater2008) and regulating emotions (Ganesalingam, Sanson, Anderson, & Yeates, Reference Ganesalingam, Sanson, Anderson and Yeates2006) have been reported years after injury. A particular type of emotion regulation involves the ability to feel emotions similar to those of another (emotional empathy), which differs from taking the perspective of others to infer how they feel (cognitive empathy; Jolliffe & Farrington, Reference Jolliffe and Farrington2006). Inability to recognize emotions of others and control emotional expression could limit emotional empathy. Although deficits in emotional empathy have been reported in adults with severe TBI (Wood & Williams, Reference Wood and Williams2008), this type of emotional processing has not yet been measured when TBI occurs during development, when the brain is undergoing rapid changes (Lenroot et al., Reference Lenroot, Gogtay, Greenstein, Wells, Wallace, Clasen and Giedd2007).
Failure of emotion regulation in typically developing children has been attributed to an imbalanced neural network involving the amygdala, a structure in the temporal lobes associated with emotion processing, and the ventromedial prefrontal cortex (vmPFC), a frontal lobe region that includes the anterior cingulate cortex (ACC) and is associated with emotion regulation (Hare et al., Reference Hare, Tottenham, Galvan, Voss, Glover and Casey2008; Somerville, Jones, & Casey, Reference Somerville, Jones and Casey2010). Connection between the amygdala and ACC reportedly increases with healthy development. Decety, Michalska, and Kinzler (Reference Decety, Michalska and Kinzler2012) found a negative relation between age and activation in the amygdaIa in healthy children and adults who completed a task that elicited empathy during functional magnetic resonance imaging (fMRI), which suggests the role of the amygdala in empathy reduces with age. The authors also found a positive relation between age and functional connectivity between the amygdala and the ACC, suggesting an increased response in the ACC across development (Decety et al., Reference Decety, Michalska and Kinzler2012).
Resting state fMRI provides a measure of the functional connectivity between structures. During fMRI scanning without task performance, low frequency spontaneous fluctuations in blood oxygen level dependent (BOLD) activity show patterns of significant correlations between regions (Biswal, Yetkin, Haughton, & Hyde, Reference Biswal, Yetkin, Haughton and Hyde1995), or functionally connected networks. The Default Mode Network (DMN) (Raichle et al., Reference Raichle, MacLeod, Snyder, Powers, Gusnard and Shulman2001) is one well-known example. BOLD signal in brain networks may be anticorrelated with that in other networks, as when the functional connectivity in the DMN decreases when functional connectivity between executive function regions increases during the performance of a cognitive task (Fox et al., Reference Fox, Snyder, Vincent, Corbetta, Van Essen and Raichle2005). Alterations to functional connectivity networks have been implicated in brain disorders (Etkin, Prater, Schatzberg, Menon, & Greicius, Reference Etkin, Prater, Schatzberg, Menon and Greicius2009; van Marle, Hermans, Qin, & Fernandez, Reference van Marle, Hermans, Qin and Fernandez2010) and injuries (Mayer, Mannell, Ling, Gasparovic, & Yeo, Reference Mayer, Mannell, Ling, Gasparovic and Yeo2011). Because functional connectivity networks are linked to neuronal function (Shmuel & Leopold, Reference Shmuel and Leopold2008; Ansari, Oghabian, & Hossein-Zadeh, Reference Ansari, Oghabian and Hossein-Zadeh2011), they may be affected by alterations in structural connectivity. Diffuse axonal injury and atrophy associated with TBI (Bigler et al., Reference Bigler, Abildskov, Wilde, McCauley, Li, Merkley and Levin2010) may impair both local and distant functional connections. However, due to diversity in characteristics of patients and methods used, information about how resting state networks adapt after TBI is still emerging. In adults with subacute mild TBI (mTBI), Mayer et al., (Reference Mayer, Mannell, Ling, Gasparovic and Yeo2011) found decreased connectivity within the DMN and hyperconnectivity between the DMN and frontoparietal network. In addition, a disruption between cingulum bundle white matter and functional connectivity of the rostral ACC (rACC) was found. The authors demonstrated that reduced connectivity within the DMN following mTBI may be related to structural abnormalities and suggested DMN disruption may be related to the increased distractibility often observed after mTBI. In adult patients, Sharp et al., (Reference Sharp, Beckmann, Greenwood, Kinnunen, Bonnelle, De Boissezon and Leech2011) reported increased functional connectivity within the DMN at least 6 months after TBI, which was related to white matter integrity. In a subsequent study of TBI patients studied 2–96 months after injury, regions of the DMN did not deactivate when subjects performed a cognitive task (Bonnelle et al., Reference Bonnelle, Ham, Leech, Kinnunen, Mehta, Greenwood and Sharp2012). Nakamura, Hillary, and Biswal (Reference Nakamura, Hillary and Biswal2009) investigated recovery of functional connectivity in adults with moderate to severe TBI between 3 and 6 months after injury. They suggested that the recovery of networks entails the strengthening of current connections rather than generation of new ones.
Although these reports provide valuable information about reconfiguration of network connectivity in adults, it is unclear how disruption to functional networks after TBI presents during development, when the brain is undergoing systematic, continuing changes (Lenroot et al., Reference Lenroot, Gogtay, Greenstein, Wells, Wallace, Clasen and Giedd2007), and whether any disruption is long lasting. Additionally, information on how TBI affects the functional connectivity of regions outside of established functional networks is needed.
Measuring functional connectivity between the rACC and amygdala may elucidate effects of TBI on a neural system responsible for controlling emotions. Frontal and temporal lobes, locations of the rACC and amygdala, respectively, are the most common sites for focal lesions after TBI in children (Graham et al., Reference Graham, Ford, Adams, Doyle, Teasdale, Lawrence and McLellan1989; Levin et al., Reference Levin, Mendelsohn, Lilly, Yeakley, Song, Scheibel and Bruce1997). In addition, children and adolescents with moderate to severe TBI show decreased cortical thickness in frontal regions that include the rACC (Wilde et al., Reference Wilde, Merkley, Bigler, Max, Schmidt, Ayoub and Levin2012) and reduced amygdala volume (Wilde et al., Reference Wilde, Bigler, Hunter, Fearing, Scheibel, Newsome and Levin2007), suggesting that functional connectivity in this emotion regulation network may be impaired.
The right amygdala has been associated with heightened response when automatically perceiving and responding to emotions (Dyck et al., Reference Dyck, Loughead, Kellermann, Boers, Gur and Mathiak2011; Glascher & Adolphs, Reference Glascher and Adolphs2003) and is involved in empathy (Decety, Michalska, & Akitsuki, Reference Decety, Michalska and Akitsuki2008). The rACC has been linked to emotional conflict resolution (Egner, Etkin, Gale, & Hirsch, Reference Egner, Etkin, Gale and Hirsch2008; Etkin, Egner, Peraza, Kandel, & Hirsch, Reference Etkin, Egner, Peraza, Kandel and Hirsch2006; Etkin, Prater, Hoeft, Menon, & Schatzberg, Reference Etkin, Prater, Hoeft, Menon and Schatzberg2010) and is activated in tasks involving impulsive choice (Hinvest, Elliott, McKie, & Anderson, Reference Hinvest, Elliott, McKie and Anderson2011). Decreased emotion regulation has been linked to impulsivity (Schreiber, Grant, & Odlaug, Reference Schreiber, Grant and Odlaug2012), and it is possible that impulsivity reported in patients after TBI (McAllister, Reference McAllister1992) may be linked to impaired functional connectivity reported in the rACC (Mayer et al., Reference Mayer, Mannell, Ling, Gasparovic and Yeo2011).
This preliminary study investigated functional connectivity between right amygdala and rACC in adolescents an average of 2 and a half years after moderate to severe TBI. We hypothesized that, relative to typically developing (TD) adolescents, functional connectivity between the two regions would be disrupted in adolescents with TBI. In addition, exploratory analyses investigated the relation of regional brain volumes and performance on empathy and impulsivity measures to the functional connectivity data.
Methods
Participants
Nine adolescents with moderate to severe TBI as defined by a post-resuscitation score of 3–12 on the Glasgow Coma Scale (GCS) (Teasdale & Jennett, Reference Teasdale and Jennett1974), or GCS score of 13–15 with associated brain pathology on computed tomography (i.e., “complicated mild”) were selected from a cohort of 75 pediatric TBI patients from a previous grant cycle (2004–2007) on the basis of availability, age 12 to 19 years, minimum 1-year post-injury, and compliant with MRI safety requirements. Patients with complicated mild injuries have worse outcome than those with mild TBI without lesions (Williams, Levin, & Eisenberg, Reference Williams, Levin and Eisenberg1990) and show no differences from patients with moderate TBI in neurobehavioral or functional performance (Kashluba, Hanks, Casey, & Millis, Reference Kashluba, Hanks, Casey and Millis2008; Williams et al., Reference Williams, Levin and Eisenberg1990). One complicated mild patient in this study (TBI1) had a GCS score of 15, measured 1 day after injury with no earlier record available. Thus, it is possible that a GCS score taken at the time of injury could be lower than 15. Excessive motion was observed in two patients, leaving seven adolescents with TBI (mean age at scanning = 17.67 years, standard deviation (SD) = 1.75, range = 14.42–19.13 years; 5 males) in the fMRI analyses (Table 1). All TBI patients had focal frontal lobe lesions on structural MRI (Table 2). Patients were recruited from hospitals in Dallas and Houston, Texas, and were studied between 1.99 and 3.63 years (mean = 2.55; SD = 0.61 years) post-injury. Ten TD adolescents served as the comparison group, although due to excessive motion in one TD subject, data from nine were included in fMRI analyses (mean age = 16.80 years; SD = 1.83; range =13.94–19.30 years; 6 males). Groups did not differ in age, gender, ethnicity, IQ, and mother's education, a proxy for socioeconomic status (all p's > .10; Tables 1 and 3). All participants were right-handed (Oldfield, Reference Oldfield1971). No child was taking psychoactive medications or had previous neurologic or psychiatric disorders. The study was completed in accordance with the Helsinki Declaration and was approved by the institutional review boards at Baylor College of Medicine and The University of Texas Southwestern Medical School at Dallas.
Features of seven patients with traumatic brain injury (TBI) and summary data for the comparison group [typically developing (TD)] adolescents
Note. AA = African American; C = Caucasian; F = female; GCS = Glasgow Coma Scale; H = Hispanic; M = male; MVA = Motor Vehicle Accident; n/a = not applicable; n = sample size; TD = typically developing; +comp = plus complications.
Pathology and associated anatomical location and size for each TBI patient
Note. Enceph = encephalomalacia; GWj = gray-white matter junction; G + W = gray matter and white matter; L = left; R = right; VP = ventriculoperitoneal.
Behavioral Measures
The Basic Empathy Scale (Jolliffe & Farrington, Reference Jolliffe and Farrington2006) is a 20-item self-report test of emotional and cognitive empathy. The Children's Impulsivity Scale (Landis & Hanten, unpublished), adapted for children from the Barratt Impulsiveness Scale-11 (Patton, Stanford, & Barratt, Reference Patton, Stanford and Barratt1995) with the authors’ permission, is a 30-item self-report scale assessing children's tendency to engage in impulsive behavior along three dimensions (motor impulsivity, non-planning, and attention). Because we predicted that the TBI adolescents would evidence decreased empathy and increased impulsivity, significant results are reported at p < .05, one-tailed.
Volumetric Measures
Brain volumetry is a measure of structure, as opposed to function, of the brain and can suggest regional abnormalities. TBI may result in degradation to neuronal cell bodies (gray matter) and/or the axons that connect them (white matter), reflected by volume loss of subcortical structures, cortical thickness reductions, or alterations in cortical folding patterns (Fischl et al., Reference Fischl, Salat, Busa, Albert, Dieterich, Haselgrove and Dale2002). T1-weighted MR images are used for automatic segmentation of brain regions into the various structures, using information such as voxel intensities, global position in the brain, and position relative to other brain structures, which are then compared to a probabilistic brain atlas to facilitate classification of each voxel in the MR image (Fischl et al., Reference Fischl, Salat, Busa, Albert, Dieterich, Haselgrove and Dale2002, Reference Fischl, van der Kouwe, Destrieux, Halgren, Segonne, Salat and Dale2004).
Functional Connectivity Measures
Functional connectivity is based on low-frequency fluctuations in the BOLD response that are postulated to follow intrinsic neuronal activity (Fox, Snyder, Vincent, & Raichle, Reference Fox, Snyder, Vincent and Raichle2007). During the measurement of functional connectivity, participants are asked to rest quietly in the scanner with their eyes open or closed. In the present study, gray and white matter volumes proximal to the rACC and right amygdala seeds were selected as regressors to understand how alterations in functional connectivity might be related to nearby tissue that may interact with blood flow (Schummers, Yu, & Sur, Reference Schummers, Yu and Sur2008).
Procedure
Subjects completed the Basic Empathy Scale and the Children's Impulsivity Scale as part of a larger battery of tasks administered on the same day or within 1 week of the scanner session. During scanning, subjects completed an fMRI task designed to measure social cognition, published elsewhere (Newsome et al., Reference Newsome, Scheibel, Hanten, Chu, Steinberg, Hunter and Levin2010), followed by resting state and structural acquisitions. During resting state acquisitions, subjects were instructed to lie still and close their eyes without falling asleep. During scanning, they were monitored via a remote camera available at one site. At both sites, subjects were asked if they fell asleep, which none did.
Acquisition Parameters
Whole brain imaging data were acquired using a multi-channel SENSE headcoil on identical 3.0 Tesla Philips Achieva scanners in Houston and Dallas. BOLD T2*-weighted single-shot gradient-echo echoplanar images (EPI) were acquired in 32 axial slices 3.75 mm thick (1.0 mm gap), 240 mm × 240 mm field of view (FOV), 64 × 64 matrix, repetition time (TR) of 1700 ms, echo time (TE) of 30 ms, 73° flip angle, and SENSE factor of 2.0. After the functional scans, a set of high-resolution T1-weighted three-dimensional (3D) turbo field echo (TFE) anatomical images was acquired in 132 axial slices of 1.0 mm thickness (no gap) with 240 mm × 240 mm FOV, 256 × 256 matrix, TR = 9.9 ms, TE = 4.6 ms, 8.0° flip angle, SENSE factor of 1.2, producing 1-mm isotropic voxels. Additional anatomical series to assess neuropathology included T2-weighted gradient echo, T2-weighted fluid attenuated inversion recovery (FLAIR), and T2-weighted gradient- and spin-echo (GRASE) sequences. Lesion volume and nature of pathology were determined by a board certified neuroradiologist (J.V.H.). Similar ranges of values for Weisskoff stability measurements [minimum 1/SNR (signal-to-noise ratio) index, peak-to-peak and root mean square (RMS) stability] (Weisskoff, Reference Weisskoff1996) taken on the day of scan indicated stability of both scanners over time.
Statistical Analysis
Behavioral Measures
Wilcoxon Rank Sums tests were performed to test group differences. Effect sizes (Cohen's d) are included to document clinically meaningful effects lacking in statistical power.
Volumetric Image Processing and Analysis
Cortical reconstruction, cortical parcellation, and subcortical segmentation of the structural MRI scans were performed using the Freesurfer neuroimage analysis suite, as described previously (Bigler et al., Reference Bigler, Abildskov, Wilde, McCauley, Li, Merkley and Levin2010; Merkley et al., Reference Merkley, Bigler, Wilde, McCauley, Hunter and Levin2008) and detailed on the Freesurfer website. Cortical white and gray matter volumes of bilateral rACC and gray matter volumes for bilateral amygdala (as the amygdala is a gray matter structure) were computed. All volumes were corrected for head size by performing analyses of covariance with total intracranial volume (TICV) (Wilde et al., Reference Wilde, Newsome, Bigler, Pertab, Merkley, Hanten and Levin2011). Wilcoxon Rank Sums tests were performed to test group differences.
Functional Connectivity Image Processing and Analysis
Functional BOLD connectivity data were spatially registered in 2D and 3D space to minimize effects of head motion, temporally interpolated to correct for slice-time acquisition differences, and de-spiked using the AFNI software package (Cox, Reference Cox1996). A regression analysis was then conducted on individual subjects’ time-series to remove potential sources of noise (physiological and machine-based) from the data based on established methodologies (Fox et al., Reference Fox, Snyder, Vincent, Corbetta, Van Essen and Raichle2005). Briefly, individual T1-weighted anatomical images were segmented into maps of white matter, gray matter, and cerebral spinal fluid (CSF); the resultant CSF and white matter masks were then used to obtain an average time-series for these tissues. Next, all six movement parameters, the region of interest (ROI) -based time-series for CSF and white matter, a constant term, and a linear term were entered into a linear regression against the extended resting state time-series. A global gray matter term was not entered into the regression to minimize likelihood of increased anticorrelations (Fox, Zhang, Snyder, & Raichle, Reference Fox, Zhang, Snyder and Raichle2009; Murphy, Birn, Handwerker, Jones, & Bandettini, Reference Murphy, Birn, Handwerker, Jones and Bandettini2009). The residual time-series data were then transformed into a standardized coordinate space (Talairach & Tournoux, Reference Talairach and Tournoux1988) in AFNI using a linear transformation.
Based on previous studies, the “seeds” for functional analyses were placed within midline rACC and right amygdala. Specifically, 12-mm spheres were generated based on voxels exhibiting maximal DMN activity in the rACC (0, 49, 9) in 42 healthy control subjects (Franco, Pritchard, Calhoun, & Mayer, Reference Franco, Pritchard, Calhoun and Mayer2009) and on voxels showing significant activation during the assessment of unfriendly behavior for the right amygdala (22.8, −10.9, −12.4; Scheibel et al., Reference Scheibel, Newsome, Wilde, McClelland, Hanten, Krawczyk and Levin2011; Schultz et al., Reference Schultz, Grelotti, Klin, Kleinman, Van der Gaag, Marois and Skudlarski2003). Resultant Pearson's correlation coefficients were then converted to Z-scores using Fisher's method, blurred using an 8-mm root-mean square Gaussian kernel, and entered into group analyses.
Random effects group analyses (two-sample t tests and multiple regression (with constant) models) were performed using Statistical Parametric Mapping software (Friston et al., Reference Friston, Holmes, Worsley, Poline, Frith and Frackowiak1995); SPM8, Wellcome Department of Cognitive Neurology, London, UK) implemented in Matlab (Mathworks Inc., Sherborn, MA). The Talairach Daemon (Lancaster et al., Reference Lancaster, Woldorff, Parsons, Liotti, Freitas, Rainey and Fox2000) was used to determine the anatomical locations and approximate Brodmann's areas (BA) of the Talairach coordinates. To isolate the effects of TBI, between group analyses are presented. For all results, significance was interpreted when voxel (height) threshold T = 2.5 and cluster threshold p < .05, false discovery rate (FDR) corrected for multiple comparisons across the whole brain. Due to small sample sizes, the p values reported are corrected for examining two tails but uncorrected for total number of regression analyses conducted. Data reduction was accomplished by including as regressors only behavioral and volume measures that showed decrements. In addition, only planned comparisons central to the main hypotheses are included, which resulted in six regression analyses.
Results
Behavioral Measures
Shown in Table 3, marginal significance and a large effect size (where Cohen's d = 0.2 is small, 0.5 is moderate, and 0.8 is large) were found in the expected direction (TBI < TD) for the BES-Emotion scale, Wilcoxon z = −1.40, p = .08, d = 1.00. The BES-C and Child's Impulsivity Scale group comparisons were nonsignificant with small to moderate effect sizes; as a result, only the BES-E scores were subsequently regressed onto functional connectivity of the rACC and amygdala. No further analyses were performed with the BES-C and Child's Impulsivity Scale data.
Mean Outcome Data for TBI and TD Groups and individual scores for TBI subjects*.
Note. Standard deviations in parentheses. WASI data for one TBI participant and BES data for one TD participant were not available. Because predictions were that the TBI group would have increased impulsivity and impaired empathy, p-values are based on one-tailed Wilcoxon Rank Sums tests. For Cohen's d, 0.2 is small, 0.5 is moderate, and 0.8 is a large effect size.
BES = Basic Empathy Scale; CIS = Children's Impulsivity Scale; n/a = not available; TBI = traumatic brain injury; TD typically developing; WASI = Wechsler Abbreviated Scale of Intelligence.
Brain Volumes
Shown in Table 4, group difference for the right amygdala volume was marginally significant with a large effect size [TD > TBI; Wilcoxon Z = −1.89; p = .059 (two-tailed), Cohen's d = 1.18], as was group difference for the left rACC white matter volume [TBI > TD; Wilcoxon Z = −1.82; p = .067 (two-tailed), Cohen's d = 1.00] volumes. All other comparisons were non-significant and had zero to moderate effect sizes.
Volumes (cc3) of amygdala and rostral anterior cingulate cortex (rACC) as a percentage of total intracranial volumeFootnote *
* Volumetric data for one TBI participant were not available. p values are based on two-tailed Wilcoxon Rank Sums tests. TBI = traumatic brain injury; TD = typically developing; ACC = anterior cingulate cortex. Standard deviations are provided in parentheses. For Cohen's d, 0.2 is small, 0.5 is moderate, and 0.8 is a large effect size.
Functional Connectivity
Is the functional connectivity of the rACC and right amygdala altered an average of 2.5 years after TBI?
Table 4 presents coordinates, cluster sizes, and probability levels of significant clusters of activation observed in the between-group functional connectivity analyses for the two ROIs. Regions listed below are voxels within a cluster (local maxima) that are 1 mm apart.
rACC seed
SPM8 analysis revealed that the TD group had greater rACC connectivity than the TBI group in two clusters (FDR-corrected cluster p = .02 for each cluster). After further correction for two tails, the FDR-adjusted probability levels were p = .04 for each of the above clusters. There were no significant clusters where TBI had greater connectivity than TD for rACC (FDR-corrected cluster p > .5301) or for amygdala (FDR corrected cluster p > .5721). Compared to the TD group, the TBI group demonstrated lower bilateral frontal connectivity between the rACC and itself (BA32) and medial (BAs 6,8,9, right-only 10), middle (BA9; right-only BA8), and superior (BAs 8,9; left-only BA6; right-only BA10) frontal gyri. Lower connectivity was also found between ACC and right fusiform gyrus (BA20) and inferior (BAs 20,21), middle (BAs 21,22,38), and superior temporal (BAs 22,38) gyri. There were no regions where the TBI subjects showed significantly greater connectivity with the rACC than the TD subjects (Figure 1a, b; Figure 2a, b; Table 5).
Significant differences between groups. Relative to the typically developing (TD) group, the traumatic brain injury (TBI) group demonstrated lower functional connectivity between the rostral anterior cingulate cortex (rACC) seed and temporal pole (a) and dorsal medial prefrontal cortex (MPFC) (b), and between the right amygdala seed and rostral and ventral MPFC (c). Activation is overlaid onto an individual subject's brain transformed into Talairach space. Left side of brain is on left side of the figure.
Fisher Z-scores of the correlation coefficients and associated standard errors between seed regions [rostral anterior cingulate cortex (rACC) or right amygdala] and the most significant voxels associated with them in the between-groups [typically developing (TD) > traumatic brain injury (TBI)] comparisons. Plotted are the correlations a) between the rACC seed and a voxel in the right temporal lobe (52, −7, −26), b) between the rACC seed and a voxel in the left medial prefrontal cortex (MPFC) (−2, 38, 46), and c) between the right amygdala seed and a voxel in the right MPFC (6, 56, 17). The coordinate points for these voxels may also be seen in Table 5, which reports coordinates and summary statistics for the between-groups (TD > TBI) comparisons.
Statistical Parametric Mapping Summary Tables for Group Analyses
Note. Height (cluster-defining) threshold: T = 2.5, degrees of freedom = 8.0; smoothness: FWHM = 14.5 14.1 12.7 (voxels); search volume = 1737604 voxels = 619.6 resolution elements (resels). Coordinate points of local maxima (voxels within a cluster) are greater than 16 mm apart. For a more complete listing of activated regions, please see the text, which reports local maxima 1 mm apart.
BA = Brodmann area; gyr = gyrus; TBI = traumatic brain injury; TD = typically developing
aProbability at the cluster level of significance after random field theory FDR correction over the whole brain search volume.
bNumber of voxels within a cluster.
cNegative values along the x-axis are defined to be in the subject's left hemisphere.
Amygdala seed
SPM8 analysis also revealed that TD had greater amygdala connectivity in one cluster (FDR-corrected cluster p = .03; p = .06 after further correction for two tails). Compared to the TD group, the TBI group demonstrated lower frontal connectivity between right amygdala and right superior frontal gyrus (BAs 9,10), bilateral ACC (BAs 24,32) and medial frontal gyrus (BAs 9,10). There were no regions where the TBI subjects showed significantly greater connectivity than the TD subjects (Figures 1c, 2c; Table 5).
Are the negative correlations in the TBI group attenuated positive correlations or anticorrelations?
Negative correlations were found between temporal and medial frontal areas in some of the TBI subjects, denoted by the subjects whose Z-scores fell below zero in Figure 2a–c. A significant negative correlation (or anticorrelation) would indicate regions whose BOLD signal increases when other regions decrease (Fox et al., Reference Fox, Snyder, Vincent, Corbetta, Van Essen and Raichle2005). For example, in the present study significant anticorrelations could reflect increased functional connectivity in the rACC or amygdala associated with decreased functional connectivity in other portions of the rACC or amygdala. Using the within-group one-sample t test model in SPM8, analysis with data from the TBI group was performed to investigate the presence of any significant anticorrelations. Only posterior areas of the brain, primarily in bilateral occipital lobe and cerebellum, showed significant negative relation with the rACC, and bilateral parietal regions showed negative relation with the right amygdala. These analyses suggest that the rACC and amygdala are not anticorrelated, and that the results from the between-groups analysis simply reflect attenuated positive correlations.
Do brain volumes relate to functional connectivity?
No significant between-group differences were observed in the relationships between rACC and right amygdala brain volumes and functional connectivity.
Do BES-E scores relate to functional connectivity?
Please see Table 6. The TD group showed a greater positive relation than the TBI group (FDR-corrected cluster p = .016; p = .032 after further correction for two tails) between BES-E and rACC connectivity in right middle frontal gyrus (BA6), precentral gyrus (BAs 3,5,6), postcentral gyrus (BAs 1,2,3,4,5,40), paracentral lobule (BA4), supramarginal gyrus (BA40), inferior parietal lobule (BA40), and superior parietal lobule (BA40) (beta = 0.02; SE = 0.01). TD and TBI adolescents did not differ on relation to right amygdala connectivity. There were no regions in which the TBI group had greater relation than the TD group between BES-E scores and either amygdala or rACC connectivity.
Statistical Parametric Mapping Summary Table for Regressions with BES-E
Note. Height (cluster-defining) threshold: T = 2.5, degrees of freedom = 12.0; Smoothness: FWHM = 16.8 16.3 15.0 (voxels); Search Volume = 1733721 voxels = 384.3 resolution elements (resels). Coordinate points of local maxima (voxels within a cluster are greater than 16 mm apart. For a more complete listing of activated regions, please see the text, which reports local maxima 1 mm apart.
BA = Brodmann area; gyr = gyrus; TBI = traumatic brain injury; TD = typically developing.
aProbability at the cluster level of significance after random field theory FDR correction over the whole brain search volume.
bNumber of voxels within a cluster.
cNegative values along the x-axis are defined to be in the subject's left hemisphere.
Discussion
Is the Functional Connectivity of the rACC and Right Amygdala Altered an Average of 2.5 Years after TBI in Adolescents?
The results of this preliminary study suggest two ways that functional connectivity of emotion regulation structures may change after TBI. First, when measuring functional connectivity from the rACC seed, the TBI group showed lower functional connectivity compared to the TD group between the rACC and frontal regions, including a portion of ACC dorsal (superior) to the rACC. The TBI group also showed lower functional connectivity with the amygdala. Apparently, the reduction in rACC functional connectivity occurs along a pathway, in regions local to the rACC as well as those more distal. Future investigations may examine the integrity of the uncinate fasciculus, a white matter tract that connects the ventral (inferior) PFC to the temporal pole. Although the uncinate fasciculus does not directly connect the rACC to the amygdala, its proximity to both structures suggests it may play a major role. Indeed, Johnson et al. (Reference Johnson, Juranek, Kramer, Prasad, Swank and Ewing-Cobbs2011) found that the integrity of the uncinate fasciculus following pediatric TBI was positively related to ability to regulate emotion. Of interest, the right uncinate fasciculus in the study by Johnson et al. (Reference Johnson, Juranek, Kramer, Prasad, Swank and Ewing-Cobbs2011) was intact, while the present study suggests impairment of the uncinate on the right side. A lower age range and more acute injuries in the Johnson et al. (Reference Johnson, Juranek, Kramer, Prasad, Swank and Ewing-Cobbs2011) study might contribute to this difference.
Second, the TBI group showed marginally reduced (p = .06) functional connectivity between the right amygdala and frontal regions, including bilateral ACC and vmPFC, implicated in emotion regulation (Decety et al., Reference Decety, Michalska and Kinzler2012; Hare et al., Reference Hare, Tottenham, Galvan, Voss, Glover and Casey2008; Sommerville et al., 2010). Moderate to severe TBI may prevent normal development of distributed functional connectivity (Fair et al., Reference Fair, Cohen, Power, Dosenbach, Church, Miezin and Petersen2009). Future experiments that directly compare adolescents and adults with similar time since injury will be important for understanding age effects of functional connectivity and their relation to emotion regulation impairments.
Findings in the between-groups analysis reflected attenuated positive correlations in the TBI group rather than anticorrelations (Fox et al., Reference Fox, Snyder, Vincent, Corbetta, Van Essen and Raichle2005). A complete reversal in the type of relation between regions that are functionally connected in healthy subjects might have been surprising given the imprecise nature of blunt force trauma.
Group difference for the impulsivity measure was not statistically significant and had only a small to moderate effect size, Cohen's d = 0.338. In a previous study of patients approximately 4 years post severe TBI, impulsivity was found only in patients with high intracranial pressure at time of injury (Slawik et al., Reference Slawik, Salmond, Taylor-Tavares, Williams, Sahakian and Tasker2009). Intracranial pressure and small sample sizes may contribute to the lack of significant impulsivity differences between groups.
The groups in this study also did not have significantly different IQ scores. Reduced IQ has been reported in children who sustained TBI at younger ages, for example, up to 9 years, but not at 10–12 years (Crowe, Catroppa, Babl, Rosenfeld, & Anderson, Reference Crowe, Catroppa, Babl, Rosenfeld and Anderson2012). It may be that the later age of injury in the adolescents in this sample (mean years = 15.1; SD = 1.8) is late enough to mitigate some effects on IQ, possibly contributing to some extent to the low power.
Is Empathy Related to Functional Connectivity of the rACC and Amygdala?
Groups did not differ on the cognitive empathy behavioral measure (BES-C, Jolliffe & Farrington, Reference Jolliffe and Farrington2006), and as a result any further analyses were not explored. Adults with TBI show emotional empathy deficits (de Sousa, McDonald, & Rushby, Reference de Sousa, McDonald and Rushby2012). In the present study, group differences in emotional empathy were marginally significant though with a large effect size. In an exploratory analysis, BES-E scores were regressed onto the connectivity pattern associated with each ROI. Relation to the right amygdala was not significant. However, a greater positive relationship between BES-E scores and rACC connectivity was found in the TD group in right posterior frontal, including somatosensory and motor regions, and nearby parietal regions, including right postcentral gyrus and inferior parietal lobule, both regions associated with emotion and empathy. Right postcentral gyrus has been linked to the recognition of emotional faces (Adolphs et al., 2000), and both it and right inferior parietal lobe have been implicated in a neural system involved in empathy. Right postcentral gyrus is associated with taking the point of view of another person, and right inferior parietal lobe is associated with taking the perspective of another person in both neutral and emotional contexts (Ruby & Decety, Reference Ruby and Decety2004). Moreover, the primary somatosensory and motor cortices have been associated with a subcategory of emotional empathy, emotional contagion (Nummenmaa, Hirvonen, Parkkola, & Hietanen, Reference Nummenmaa, Hirvonen, Parkkola and Hietanen2008). Emotional contagion involves the mimicking of facial and other movements of another person, resulting in experiencing similar feelings (Hatfield, Cacioppo, & Rapson, Reference Hatfield, Cacioppo and Rapson1993). Any reduced empathy experienced by patients after TBI may be associated with reduced connectivity between the rACC and these regions.
Are Volumes of the rACC and Right Amygdala Related to the Functional Connectivity of the rACC and Amygdala?
Groups did not differ in rACC volume, so further analyses were not performed. Groups differed marginally on the right amygdala, but with a large effect size, consistent with previous findings reported by our group (Wilde et al., Reference Wilde, Bigler, Hunter, Fearing, Scheibel, Newsome and Levin2007). Whereas the effect size of group differences was large for the right hemisphere, it was very small for the left (Cohen's d = 0.01). Amygdala volume asymmetries have been observed in major depression, a disorder with noted emotion dysregulation, where right amygdala was reduced compared to left (Mervaala et al., Reference Mervaala, Fohr, Kononen, Valkonen-Korhonen, Vainio, Partanen and Lehtonen2000; Xia et al., Reference Xia, Chen, Zhou, Zhang, Yang, Xia and Wang2004). This is in contrast with a lateralization effect noted in healthy adults, where larger right amygdala volume compared to left has been reported for right-handed individuals (Szabo, Xiong, Lancaster, Rainey, & Fox, Reference Szabo, Xiong, Lancaster, Rainey and Fox2001). The current findings suggest that reduced emotional empathy and emotion regulation following TBI may be related to right amygdala volume reduction.
In an exploratory analysis to investigate relations between the volume of the right amygdala and functional connectivity, group differences were nonsignificant. Given that loss in gray matter may entail a reduction in neurons, replication with larger sample sizes would be necessary before completely ruling out the role of atrophy in functional connectivity alteration.
How Might the Lesions in the TBI Group Affect the Findings?
Corbetta (Reference Corbetta2012) described how locations of focal lesions can be very different from locations where associated functional aberrations are found, e.g., ventral frontal and temporoparietal lesions in stroke patients with spatial neglect (inability to attend to one side of space) have been associated with aberrant functional connectivity in dorsal frontal and parietal regions. To contribute to the uncertainty of how focal lesions and functional connectivity are related, the TBI sample in this study is representative of the TBI population in that it has heterogeneous lesions. Although frontal lesions occurred in six out of seven patients, the locations varied. Because the frontal lobes share connections with many parts of the brain (e.g., fronto-temporal, fronto-parietal, fronto-occipital, fronto-cerebellar networks), lesions may potentially have different effects on different networks, with the location of pathology within a network potentially having still further implications for aberration in functional connectivity. In an investigation of the effects of structural lesions on functional connectivity using a computational model, Alstott, Breakspear, Hagmann, Cammoun, and Sporns (Reference Alstott, Breakspear, Hagmann, Cammoun and Sporns2009) found that lesions to midline structures (including frontal), in the model resulted in extensive functional disruption between frontal, temporal, and parietal regions in both hemispheres. When lesions were made in lateral areas, the effects on functional connectivity tended to remain in the same hemisphere. Given that many of our TBI subjects had bilateral frontal lesions, we infer that extensive disruption in the functional connectivity occurred between frontal, temporal, and parietal regions. The disrupted functional connectivity between the rACC and right amygdala may be, in part, a product of the lesions to the frontal lobes and, in part, atrophy found in the right amygdala, which may have lateralized effects (Alstott et al., Reference Alstott, Breakspear, Hagmann, Cammoun and Sporns2009) and affect functional connectivity on the right side.
Limitations
Limitations of this preliminary study include small sample sizes and should be viewed with caution. We also acknowledge that the findings may not be completely generalizable to adolescents who are unable to remain still in an MRI machine and may have diminished cognitive and/or motor control, which could be associated with differences in brain volume and connectivity. Regional brain volume is only one factor that may affect functional connectivity, and may itself reflect numerous physiological processes associated with glial cells and myelinated axons as well as developmental changes following TBI that have not yet been well characterized. It may be that a subset of those physiological processes do not have a large effect on functional connectivity and override any processes that are influential. Furthermore, future studies with Common Data Elements measures would facilitate a comparison of TBI adolescents to normative samples, which was not possible in our study because the behavioral measures we used were not standardized.
Studies with larger sample sizes will be important for investigating the relation of lesion characteristics on functional connectivity, and studies with a wide range of post-injury intervals will be important to investigate the role that time since injury may play in functional connectivity alterations. Furthermore, although correlative analyses of functional connections provide valuable information, analysis methods involving causal modeling would test the direction in which regions influence each other, adding another dimension to typifying how functional connections can be altered.
Conclusions
In this preliminary study, functional connectivity was altered on average 2 and a half years after moderate to severe TBI in adolescents. Reductions in functional connectivity were found between the right amygdala and medial prefrontal cortex, including the rACC, an area implicated in the ability to regulate emotions. Moderate to severe TBI may prevent normal development of distributed functional connectivity. Our results may suggest that adolescents are more vulnerable than adults to disruption of emotion regulation connectivity in the years following moderate to severe TBI. Thus, rehabilitation focused on regulating emotional response may be of particular relevance in adolescents.
Acknowledgments
This work was supported by NINDS grant R01-NS21889 “Neurobehavioral outcome of head injury in children,” Harvey S. Levin, PI. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. There are no potential conflicts of interest, including financial, between any of the authors and the work presented in this manuscript. We are grateful to the youth and their families for taking part in the research, to the South Central Mental Illness Research, Education, and Clinical Center (MIRECC) for providing facilities that allowed analyses of the fMRI data, to the Editor, Associate Editor, and to three anonymous reviewers who provided valuable feedback. Zachary Dorsey and Isaiah Tate aided in the preparation of the manuscript.
